From my perspective, the point of raising the example of natural selection is that it debunks the naive belief that if in general an outer optimizer trains on reward function, it gets an inner optimizer that pursues that reward function OOD. Saying "But SGD is first-order and
RESEARCH
-
Logical fallacy: Equating two misunderstood phenomena
By
–
Up there with "I don't understand consciousness; I don't understand quantum physics; therefore consciousness is quantum"
-
Anthropomorphizing AI: Avoiding Precise Definitions of Model Faculties
By
–
Step 1: anthropomorphize AI models by claiming they possess various human faculties. Stay away from any precise definition of the faculties, and assert that the claim is self-evident because AI models can show human-like outputs in certain situations. Step 2: when evidence of
-
Consciousness in humans versus AI systems: evidence and distinction
By
–
New twist on this: "there is no evidence that humans are conscious, either!" (We are inundated with evidence that humans and most animals around us are sentient and conscious, unlike hashtables, parametric curves, and related data retrieval systems)
-
Lack of rigor and knowledge in AI field criticized
By
–
“I have this idea where 6 – 3 = 72 and I am pretty sure nobody has thought of it, so you mind if I call it math?” This is person has zero knowledge and has made zero effort. Doesn’t even know the difference between transformative and derivative and still feels qualified to
-
Direct Preference Optimization: Language Models as Reward Models
By
–
Direct Preference Optimization: Your Language Model is Secretly a Reward Model Rafailov et al.: https://
arxiv.org/abs/2305.18290 #ArtificialIntelligence #DeepLearning #MachineLearning -
Augmented Intelligence vs IA: Terminology and Force Multiplication
By
–
i usually cringe and naming discussions but i dig it a lot. am coming from a perception and AR/VR background here, so i've always been partial to the term of "augmented intelligence" – but it lacks the force multiplication aspect that IA nicely captures.
-
Direct Preference Optimization: Language Model as Reward Model
By
–
Direct Preference Optimization: Your Language Model is Secretly a Reward Model Rafailov et al.: https://
arxiv.org/abs/2305.18290 #ArtificialIntelligence #DeepLearning #MachineLearning -
DeepMind vs OpenAI: Betting on Demis’ Leadership
By
–
In the race between DeepMind and OpenAI my bet is still on Demis.
-
Critique du raisonnement attribuant la conscience aux modèles d’IA
By
–
"Humans are conscious; this big curve fitted on tons of human-generated outputs can reproduce human-like behavior in some cases; therefore this big curve is conscious" has got to be some of the most mindless, most hubristic reasoning I've ever seen.